J. S. Af,. Inst. Min. Metal/., vol. 86, no. 4.

Apr. 1986. pp. 113-124.

Abrasive and impactive wear of grinding

balls in rotary mills
by LA. VERMEULEN* and D.D. HOWATt

SYNOPSIS
New distribution functions are derived to describe the size distribution of grinding elements in ball mills. The
formulations are based on the assumption that abrasive as well as impactive interactions occur during ball milling-an
assumption that is supported by a large body of experimental evidence. It is shown that the functions can be used
in the estimation of the magnitudes of abrasive and impactive components in the total rate of ball wear.
The relative magnitudes of the wear components provide a basis for optimization of the chemical and metallurgical
properties of the balls in a given milling situation. However, it is contended that these quantities are also useful
indicators of the relative intensities of the abrasive and impactive interactions that are operative in the size reduction of mineral particles within ball mills. These quantities can be determined for any industrial ball mill, and their
magnitudes provide practical guidelines for mill operation.
The theory is used in an analysis of samples of ball charges from two industrial ball mills, and it is also applied
to all the data on ball-size distributions found in the literature. The qualitative correlation between the calculated
values of the wear components and the reported operating conditions is good for a variety of industrial ball mills,
in conformity with the hypothesis that the relative magnitudes of the wear components are related to the milling
.
conditions.
SAMEVATTING
Daar word nuwe verdelingsfunksies
afgelei om die grootteverdeling
van maalelemente
in balmeule te beskryf.
Die formulerings
word gegrond op die aanname dat daar sowel skuur- as slagwisselwerking
tydens balmaling
getuienis gestaaf word. Daar word getoon
plaasvind-'n
aanname wat deur 'n groot hoeveelheid eksperimentele
dat die funksies gebruik kan word om die grootte van die skuur- en slagkomponente in die totale tempo van balslytasie
te raam.
Die relatiewe grootte van die slytkomponente
verskaf 'n grondslag vir die optimering van die chemiese en
metallurgiese eienskappe van die balle in n gegewe maalsituasie. Daar word egter aangevoer dat hierdie groothede
ook nuttige aanwysers is van die relatiewe intensiteit van die skuur- en slagwisselwerking
wat plaasvind tydens
die verkleining van mineraalpartikels
in balmeule. Hierdie groothede kan vir enige industriele balmeul bepaal word
en verskaf praktiese riglyne vir meulbedryf.
Die teorie word gebruik om monsters van balladings wat uit twee industriele balmeule verkry is, te ontleed en
word toegepas op al die data oor balgrootteverdelings
wat in die literatuur gevind is. Die kwalitatiewe korrelasie
tussen die berekende waardes van die slytkomponente
en die gerapporteerde
bedryfstoestande
is goed vir 'n
verskeidenheid industriele balmeule in ooreenstemming
met die hipotese dat die relatiewe groothede van die slytkomponente met die maaltoestande
verband hou.

Introduction
Ball milling has been employed for more than a hundred years in the fine grinding of ores, coal, cement, and
other materials. The world's consumption of grinding
balls used in this way is about 500 kt a year, and the consumption of balls per ton of material milled varies widely. Ball consumption constitutes a significant proportion
of the costs of fine grinding, rising markedly for hard
and abrasive ores. In attempts to reduce ball consumption and the high cost of fine grinding, many studies have
been directed to the analysis of factors causing ball wear.
These studies have almost invariably attributed all the
wear to one of the two wear mechanisms that operate in
ball milling: abrasive wear and imp active wear.
That abrasion plays a large part in all fine-grinding
operations has long been known and acknowledged. Over
forty years ago, Prentice' reported tests aimed at providing evidence that all the wear of grinding balls is caused
by abrasion. Much additional evidence has since been
. SpecialistScientist,Council for MineralTechnology(Mintek),
Private Bag X3015, Randburg, 2125 Transvaal.

Emeritus Professor, University of the Witwatersrand; now at Mintek

(see address above).

@ The South African Institute of Mining and Metallurgy, 1986.
SA ISSN 0038-223X/ $3.00 + 0.00. Paper received 8th July, 1985.

quoted2 to show the significance of three-body abrasive

interactions in fine grinding, which result in the size
reduction of minerals and in the wear of grinding media.
The wear of grinding balls by impace in a mill is not
nearly as evident nor as widely accepted. Nevertheless,
there is ample proof that balls are projected into flight
and collide with the en masse charge and occasionally
with the mill lining. Numerous photographic studies, the
recent techniques of instrumented bolts4 in mills, and
the results of tests on balls fitted with accelerometers5
have all produced irrefutable evidence of impact.
During impactive processes, the rate of ball wear,
which is the rate of mass loss by the balls, is proportional
to the ball mass, Le. the cube of the ball diameter,
whereas the rate of ball wear during abrasive processes
is proportional to the surface area' of the balls, or the
square of the ball diameter.
The general idea of combined wear, in which
mechanisms of both abrasive and imp active wear are
operative, can be attributed to Bond6. His work in this
area can be summarized,,7 by the following empirical
relationship between the rate of ball wear, - dm/dT, and
the ball diameter, X:

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

APRIL 1986

113

dm
dT

k' Xq,

(1)

where m is the mass of a ball, T the amount of material

milled, and ,k I and q are constants. The value of the exponent q, which by hypothesis should lie in the range
2 ~ q ~ 3, provides an indication of the relative in tensities of the two components in the rate of ball wear.
In discussions of the formula given in equation (1),
Bond6,s, Hukki9, and Taggart' emphasized the role of
mill speed and pointed out that, at low speeds, say in the
neighbourhood of 60 per cent of critical speed, cascading
will be a prominent feature of the charge motion and q
will be very nearly equal to 2; at higher speeds, say 85
per cent of critical speed and higher, cataracting will be
a prominent feature and q will be nearly equal to 3.
The idea that the abrasive and impactive components
in ball wear are functions not only of the ball material
but also of the mill-operating variables is therefore well
established in the literature. As ball wear implies size
reduction during the residence of balls in a mill, it can
be expected that the size distribution of balls in a mill
will be related to the relative intensities of the two wear
mechanisms. These values can be of practical significance
in the selection of the best chemical and mechanical
characteristics of balls for a given milling operation; they
can also be used to reveal the conditions in the mill in
which the balls are operating as the grinding medium. For
this reason, the present investigation is based upon data
from measurements of ball-size distributions, and includes the derivation of expressions to describe a steadystate ball-size distribution under conditions of abrasive
and impactive wear, and the determination of the relative
magnitudes of the two components in the rate of ball wear
under defined conditions of milling.
Relationships between Wear Mechanisms and Ball-size
Distribution
Only ball wear is considered here. The production of
smaller 'balls' by the fracture of larger ones is beyond
the scope of the present work.
Qualitative and quantitative information relating to the
relative intensities of the two wear components can be
obtained from studies of ball-size distributions by the
establishment of a relationship between the number density, ,,(X), of balls of diameter X and the rate of size
reduction ofthese balls, - dX/dT(millimetre
per ton of
material milled). The number density function, which is
often called the' frequency distribution', is defined by the
relation
dN = ,,(X)dX,

(2)

where dN is the number of balls in the mill whose

diameters are in the size interval X to X + dX, and X
is within the range Xo ~ X ~ Xmax'
From equation (2) it can be shown that the mass of
this aggregate of balls is
dM = .~ P7rX' ,,(X)dX,
6
114

APRIL 1986

(3)
JOURNAL

where p is the density of the ball material.

Ball-size distribution in a given mill is determined conventionally as follows: the entire ball charge is removed
from the mill and screened into a number of size intervals; the number and mass in each size interval are then
determined. In this way, experimental values are obtained for the distribution functions n(X) and m(X), which
are, respectively, the cumulative number fraction and the
cumulative mass fraction of balls passing size X. These
distribution functions are related theoretically to the
number density function, ,,(X), according to the following formulae:
x

,,(X)dX

n(X)

Xo

(4)

xmax

(X)dX

Xo

"

and

j XJ"

m(X)

(X)dX

Xo
-

(5)

xmax
~

XJ,,(X)dX

Xo

The charge of a ball mill has been shown1,J,9to tend

to a sta,bilized condition after the mill has been operating
under steady-state conditions for a sufficiently long time.
Steady state implies a constant feed rate of ore to a mill
and a steady rate of addition of top-size balls to compensate for the depletion of the ball charge by wear. The
time required for the stabilized condition to be achieved
is about
-3M
,
(dM/d1) (dT/dt)
where M is the total mass of balls within the mill, dM/ dT
is the rate of ball consumption (kilograms per ton of ore
fed to the mill), dT/ dt is the milling rate, and the negative
sign appears because the mass of balls decreases during
the milling operation. It is assumed that, in the stabilized condition, the ball-size distribution in a mill does not
vary!.J; namely, that during operation the number and
mass of balls in any specific size interval remain constant,
although these quantities may vary substantially from one
size interval to another. The number density, ,,(X), is then
related to the rate of ball wear by a consideration of the
number of balls in a size interval, as follows.
If Nj is the number of balls in the mill whose
diameters are in the size range Xi to ~, where Xo ~ Xi
< ~ ~ Xmax' then from equation (2)
Xi

N, =

j ,,(X)dX.

(6)

xi
If ~ is constant, its derivative with respect to the
amount of material milled, T, is zero, Le.

OF THE SOUTH

AFRICAN

INSTITUTE

OF MINING AND METALLURGY

Xj

dN

d
- v (X)dX
'
dT
dT Xi

V(X)

dX

'

dT

dX,
V(X) = O.
dT

(7)

Since equation (7) holds good for any X < X within

the range Xo to Xmax' it follows that
-

dX

veX)

is constant.

(8)

dT

From a consideration of the largest size group, it can

be shown!! that the constant in equation (8) is n', the
number of top-size balls of diameter X
that are added to the mill in the time required for th~ mill to grind
a unit mass of new feed. Equation (8) can therefore be
written in the form

where the subscript C denotes combined wear.

Table I gives expressions for Nco the total number of
balls in the mill, and Mco their total mass, and for the
functions nc(X,q) and mc(X,q), which are number and
mass distribution functions respectively. These expressions were obtained simply by the substitution of equation (11) into equations (2) to (5), followed by the required
integrations. Table I also shows the forms to which the
expressions reduce under hypothetical conditions of purely abrasive wear! and purely impactive wearJ, under
which the exponent q has the limiting values of 2 and 3
respectively.
As an example of the applicability of the theory, the
number distribution function, nc(X,q) (equation 14 in
Table I), was fitted to measured values7 of the ball-size
distribution in the charge of a primary ball mill at
Marievale (Table A-I in the Addendum). The fitting procedure yielded the formula
XO,J6

nc(X,q)

X~;~6

(X)

~.

(9)

dX/dT

Equation (9) shows that, in a stabilized mill, the

number density of balls of size X is inversely proportional
to the rate at which their diameters are being reduced.
This relation, together with equations (2) to (5), can
therefore be used to supply information on the relative
intensities of the wear mechanisms that are operative in
any given stabilized mill.
The exponent q in the empirical formulation of the rate
of ball wear given in equation (1) provides important,
although only qualitative, information on the relative intensities of the rates of abrasive and impactive wear in
ball milling. This is because a value of q between 2 and
3 represents combined wear, in which the mechanisms of
both abrasive and imp active wear are operative. The
determination of q from measurements of the ball-size
distribution requires the derivation of the number density function and the distribution functions appropriate to
this formulation, followed by fitting of the derived
distribution functions to the measured values of the
number and mass fractions of balls passing size X.
From Bond's formulation of the rate of ball wear ,
which is expressed by
-

dm
= k' Xq,
dT

(1)

the following is obtained:

dX
dT

= kXq-2,

(10)

where k = Zk' / p7f. Substitution of equation (10) into

equation (9) then yields the following expression for the
number density function:
vc(X,q) = -, n'
kXQ-2
JOURNAL

OF THE SOUTH

(11)

XoOJ6

(16)

XoO,J6

where Xo = 1,0 inch and Xmax = 4,0 inches. For purposes of comparison, the respective number distribution
functions nA(X) and n!(X) for purely abrasive and purely impactive wear (Table I), with the same values for X
and Xmax' are also depicted in Fig. 1. The functio~
nc(X,q) for combined wear gives the best description of
the measured ball-size distribution. Equations (14) and
(16) show that q, as determined from the numberdistribution function, is
q(nc) = 3 - 0,36 = 2,64.
A similar analysis, in which the mass distribution function for combined wear mc(X,q) and the measured mass
distribution were used, yielded
q(mc) = 2,28.
Despite a discrepancy of about 15 per cent between the
two estimates of q, the results suggest fairly strongly that
the balls in the mill were subjected to a significant impactive component.
It is clear that exponent q is a number, and that the
determination of its value provides only a qualitative indication of the relative intensities of abrasive and impactive wear in a given mill.
Bond's empirical relation, which was shown in equation (1), can be replaced by a more detailed analysis that
leads to the determination of the magnitudes of the two
components in the rate of ball wear. The rates of abrasive
and impactive wear are superposed in the following manner: during abrasion the rate of ball wear is proportional
to the surface area of a ball!, Le.
dm
ex m2/J;
- dT
during impact the rate of ball wear is proportional to the
mass of a ballJ, Le.

- -dm

ex

m.

dT

AFRICAN

INSTITUTE

OF MINING AND METALLURGY

APRIL 1986

115

TABLE I
EXPRESSIONS TO CHARACIERIZE A STABILIZED BALL-SIZE DISTRIBUTION UNDER COMBINED WEAR AND UNDER THE HYPOTHETICAL
CONDITIONSOF PURELY ABRASIVE AND PURELY IMPACTIVE WEAR ACCORDING TO THE BOND FORMULATION OF THE RATE OF BALL WEAR

Formula

for wear rate

-~;OCxq.............

Number

density function

vc(X,q)

Number
mill

of balls in the

Nc(q)

(1)

n'
kXq-2

. . ..

X:-q)
n' pJt(X;;;.q -

dm
dT

vA(X) = ~
k

(12)

NA = k n' (Xmax-

n'pJt
4
MA - 24j{(Xmax-X.)

Number distribution
function of balls
smaller than size X

nc(X,q) =
q
x~xq - X;

. . . . . . (14)t

X-X.
nA(X) =
Xmax- X.

Mass distribution
function of balls
smaller than size X

x6-q - X:-q
mc(X,q) =
q
X.:..q- X:

. . . . . . (15)t

mA(X) =

x3-q

- X.q-3

oc m

n'
N, = T(lnXmax-lnX.)

X.)

Mc(q) =

. , . . (13)

dm
dT

v,(X) = n'
kX

Mass of balls in the

mill

6k(6 - q)

(limiting form as q ~ 3)

oc m%

(11)
"""""

n'(X,;;;;.q - x;-q)
k(3 - q)

Purely impactive wear3

Purely abrasive wear7

(limitingformasq~2)

General expression for combined

(abrasive and impactive) wear'

Quantity

X4 - X.4
- X.
X':'"

M, -

pit
n'rsi(
3
( Xmax

n.(X)

InX-lnX.
= InXmax-lnX.

m,(X) =

- X.3 )

X3 - X;
X~ - X;

'The present investigation

Subscript C denotes combined wear, Subscript A denotes purely abrasive wear, Subscript I denotes purely impactive wear
t Austin and Klimpel20recently also derived these functions by different methods

Xm"
0
n,(X)

m..

nA(X)
ndX,q)

~4.0in
~ Experimental data'
~ Number fraction for purely impacUve wear
Number fraction
~ Number fraction
q ~ 2,64
when

for purely abrasive wear

for combined wear,

,./'{/!
...../" 1/

0,8

0,6

'<'

,/

..

"
B",
~~
1:
" .,

'.

WI(X)

ap7rX3

(18)

6
and

WA(X)

1
1

../~' /f

(3(! p7rXY3
6

(19)

~"/~c/~/ /
04
'

]
U
"

/;:;.

E ,=

'~

.i

The two terms on the right-hand side of equation (17)

can be expressed in terms of ball size as

.l/
...... 11
./ /
..../ 1/

where a and (3 are constants.

~ 1.0in

x"

1,0

../

/ /

where WA(X) and W\(X) denote the respective rates of

abrasive and impactive wear on balls of size X within the
range Xo ~ X ~ Xmax'
A number density function based upon this formulation of the rate of ball wear can be obtained easily.
Manipulaiion of equation (18) yields

//
i~/
; 1

0,2

//
i/

//
I

'I
1

dX

Ball size X. in

d!

Fig. 1-Comparison of the measured size distributions of balls

in the charge of the no. 1 primary ball millat Marievale with those
calculated for three types of wear

1/3 a(X + A),

(20)

where

A=

(3
a(p7r/6)

113

(21)

Therefore combined wear should, to first order, be given

bi,I2
dm
- -

dT

116

Dim

+ flm2i\

APRIL 1986

,..,..,

(17)
JOURNAL

The number density function, obtained by the substitution of equation (20) into equation (9), is

OF THE SOUTH

AFRICAN

INSTITUTE

OF MINING AND METALLURGY

TABLE 11
EXPRESSIONSTO CHARACfERIZE A STABILIZED BALL-SIZEDISTRIBUTION UNDER CONDITIONS OF COMBINED WEAR
Quantity

General

Formula

for wear rate

dm
- - dT

Number

density function

vc(X,I.) =

Number

of balls in the mill

Nc(l.) = 3n'
a- In

distribution

Mass distribution

(abrasive

and impactive)

wear'

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..

( Xo ++AA)

n'pJ[
2(\

Xmax

-X 0
[ ...L(x>
3
max

In (X + A) -In

9~

LI.(X2
2

max

_X2)+1.2(X
0

max

( Xmax+I.
+ A )J

-X)-I.>ln
0

.. . . . ..

mc(X,I.)=

function

(Xo + A)

(X> - X;) -

...L(X3max -X3)0

upon the superposition

of abrasive and impactive

(24)

Xo

(25)

nc(X,I.) =
In(Xmax+I.)-ln(Xo+l.)

function

(17)
(22)

'Based

for combined

(\(ln~l.)

M c (A) =

Mass of balls in the mill

Number

2;

= (\m+~m;>

expression

I.(X2 - Xo2)

+ 1.2(X

LI.(X2max _X2)+1.2(X
0

- Xo) - 1.>ln

max

( :.:~ )

( Xmax+l.
)
+A

-X)-I.>ln
0

. . . . . . . . ..

(26)

Xo

wear rates

3n '
.
(22)
a(X + A)
[his function is expressed as a function of A, because the
latter has important physical significance within the present context, which will be discussed further.
Expressions equivalent to those listed in Table I, derived by the subi'titution of equation (22) into equations (2)
to (5), are shown in Table 11. In addition to these
equations-(23) to (26)-expressions can be derived for
the cumulative number and cumulative mass of balls
smaller or larger than a given size, the number and mass
of balls in a size interval, the mean ball diameter, the surface area of the ball charge, and so on.
The parameter A in the expressions above is proportional to {31a so that, if {3tends to zero, then Aalso tends
to zero. The expressions in Table 11 then reduce to the
corresponding expressions that are applicable under
hypothetical conditions of purely imp active wear (Table
I). On the other hand, if {3is finite and a tends to zero,
then A tends to infinity. The expressions in Table 11then
reduce to the corresponding expressions that are valid
under hypothetical conditions of purely abrasive wear
(Table I).
vc(X, A) =

The physical meaning of A becomes apparent from

equations (18), (19), and (21) when the late of abrasive
wear is equated to that of impactive wear. This shows
that A represents a ball size for which, under the given
milling conditions, the ratio of abrasive to impactive wear
is unity. If AIXmax > 1, then all the balls in the mill experience primarily abrasive wear, Le. the action of the
mill is primarily abrasive, but there is a finite impactive
component as well. If AIXmax< 1, then all the balls in
the mill with diameters larger than Aexperience primarily imp active wear, and all thos:: with diameters less than
A undergo primarily abrasive wear. The action of the mill
is then clearly such that balls are subject to both wear
mechanisms, but the impactive component is now
enhanced relative to the previous case, in which AIXmax
was greater than 1.
Equations (18) and (19) give the respective rates of
abrasive and impactive wear on balls of size X. Summation over all the balls in the mill gives the values for
!VA(charge) and W,(charge), the respective rates of
abrasive and impactive wear on the whole ball charge,
as shown in equations (27) and (28).

Xmax

WA(charge) =

WA(X)V(X,A)dX

Xo

= Yz n'p1TA [Y2(X~ax - X2)0

A(X max. - X)0

+ A2ln

Xmax + A
Xo

(27)

+ AJ

Similarly,
W,(charge) = Yz n'p1T
JOURNAL

OF THE SOUTH

1/3(X~ax

AFRICAN

- X~)

INSTITUTE

Y2A(X~ax - X~)

+ A2(Xmax- Xo)

OF MINING AND METALLURGY

AJlnXmaX+ A

(28)

1986

117

Xo + A J
APRIL

Hence, the ratio 0 f the rates of abrasive to impactive wear on the ball charge is
V2A(X~ax - X~)
R Al charge)

- A2(Xmax- XJ

- X~)

V2A(X~ax - X~)

+ A2(Xmax- Xo)

Equation (29) shows that the ratio RAI(charge) is independent of lX, (3, and the rate of ball consumption. It
is a function only of A, Xo, and Xmax' The quantity A
can therefore be used to express quantitatively the relative
magnitudes of the rates of abrasive and impactive wear
that are operative on the whole ball charge. It is contended that not only are these quantities related to ball wear,
but they are also important indicators of the mechanisms
of size reduction of mineral particles in a given milling
situation. Hence, their calculated magnitudes should correlate with the milling conditions, e.g. mill diameter, mill
speed, liner configuration, pulp density, ball size.
Furthermore, A can also be related to the exponent q
in the Bond formulation of ball wear. The relationship
is found by the use of equations (12) and (13) in Table
I in conjunction with equations (23) and (24) in Table
11, to yield equation (30):
- X~)

V2A(X~ax - X~)
1n(Xmax + A)

+ A(Xmax - Xo)

APRIL 1986

JOURNAL

.(29)

AJln Xmax + A
Xo + A

mass unless all the grinding elements are truly spherical.

The degree of uncertainty to be associated with theoretical
assessments of the abrasive and imp active components
in ball wear can be estimated by the use of this difference
between the measured and the calculated ball masses.
The Nature of Ball Wear in Some Industrial Mills
The determination of the ball-size distribution in an
industrial mill is a lengthy process involving disruption
of the milling process and possibly some loss of production. Although the alternative procedure, Le. sampling
of the ball charge, introduces an element of uncertainty,
it was carried out in two industrial mills-a 9 ft by 10 ft
ball mill at the Libanon Gold Mine and an 8 ft by 8 ft
ball mill at the Marikana Mine of Western Platinum
Limited-and the data obtained were analysed. In addition, the literature was searched for all the relevant data
-

AJln Xmax + A
Xo + A

1n(Xo + A)

This relationship shows that, if the value of A is known,

q can be determined, and vice versa.
The new distribution functions ne(X,q), me(X,q),
ne(X,A), and me(X,A) shown in Tables I and 11
therefore provide various methods for the determination
of A, and hence of the rates of abrasive and impactive
wear during ball milling.
In mathematical integrations over a range of ball sizes,
the quantities k, lX,and {3in the two formulations of ball
wear were regarded as constants, this assumption being
common to the present work and all previously reported
investigationsl,J,zo. These quantities are essentially shape
factors, implying that the derived expressions are applicable strictly only to size distributions in which the
grinding elements remain spherical. Examination of the
ball charge in any industrial mill will show that a substantial proportion of the smaller grinding elements are not
spherical, and some degree of error therefore arises when
the derived distribution functions are applied to ball
charges.
The complete evaluation of these errors is difficult, but
the constant-shape approximation is good for balls larger
than one-third of the size of top-size balls because these
balls are fairly well rounded. The smaller, non-spherical
elements usually constitute less than one-third of the total
number and a much smaller proportion of the total mass.
The non-sphericity factor can be taken into account if
the total mass is calculated according to equations (13)
or (24) and the results are compared with the measured
total mass of the charge. The value of the calculated mass
will always be smaller than the value of the measured
118

Xmax + A
Xo + A

l/3(X~ax

l/3(X~ax

+ AJln

(3 (6 -

q) (X~:xq -

x,~-q)

q) (X~:xq

xt'1)

(30)

that could be used in testing the validity of the theory

of combined wear.
Data reported by PrenticeI for the whole ball charge
in a 6 ~-'2ft by 12 ft mill at Blyvooruitsig Gold Mine included the numbers and masses of balls in every V2-inch
size interval between 1 and 3 inches, the total number and
total mass of the balls, and the rate of ball consumption,
which was equivalent to 300 top-size balls per day. The
mill had been operating under steady-state conditions for
8 months, and it seems a safe assumption that the charge
was stabilized, since about 9,5 times the total ball load
had been consumed in that period. The throughput of
the mill was not reported, but it has been suggestedlJ
that it was about 450 t/ d, giving a rate of ball consumption of 0,667 balls or 1,17 kg of balls per ton milled. The
original datal are reported in Table A-2 of the Addendum. Such comprehensive data provide a basis for illustrating the application of the various formulae in the
calculation of the rates of abrasive and impactive wear
under given milling conditions.
The cumulative forms of Prentice' s data are shown by
the points in Figs. 2(a) to 2(d). The distribution functions
ne(X,A) and me(X,A), shown in equations (25) and (26),
were adapted to the data. The fitting procedures, with
Xo equal to 0,6 inch, gave the following values:
ACne) = 8,1 inches = 206 mm
ACme) = 8,9 inches = 226 mm.
In (a) and (b) of Fig. 2, the cumulative forms of the
Blyvooruitsig valuesl are compared with graphs of the

OF THE SOUTH

AFRICAN

INSTITUTE

OF MINING AND METALLURGY

f
'f

(a) -=Distribution
function
nc(X,"A) with "A=8,1 inches
@)
0,8
= Prentice's experimental
datal

R
t::
'"'-

(c)-

= Distribution function nc(X,q)

with q=2,16
@)
= Prentice's experimental
datal

~
';:: i!:! 0,6
;;j';;;
....

co

~ ,~
';::
.!So.~

0,4

='
S

u='

0,2

R
'"'-

(b)

-=

Distribution

function

with "A= 8,9 inches

@)
= Prentice's experimental

0,8

mc(X,"A)

(d)-=Distribution
function mc(X,q)
with q = 2,20
<!) = Prentice's experimental data I

datal

EO

d"
,g

0,6

gf
e'~
... 0:1

0,4

t;i
"3
S

0,2

;;j
i!:!
<Z::';;;
~
,~

0.

u='

Fig. 2-Comparison

XO,S4
=

Ball size X, in

of measured and calculated size distribution of balls in the charge of the ball mill at Blyvooruitsig Gold Mine

functions nc(X,;\) and mc(X,;\) plotted from the above

values of the parameters. It can be seen that the curves
are good descriptions of the measured values. The
discrepancy between the two values of ;\ is less than 10
per cent, and the mean value (216 mm) is nearly three
times larger than that of the top-size balls fed to the mill.
This indicates that, although abrasive processes were
dominant in the mill, there was a finite imp active component as well.
The distribution functions nc(X,q) and mc(X,q) as
shown in equations (14) and (15) respectively (Table I)
were also fitted to Prentice's data. The fitting procedures
yielded the formulae
nc(X,q)

Ball size X, in

3,84

- 0,6,84

0,6O,S4

and
X3,so
- 0 63,so
mc(X,q) =
'.
33,so 0,63,so
-

In (c) and (d) of Fig. 2, the cumulative forms of the

Blyvooruitsig valuesl are compared with graphs of these
formulae for nc(X,q) and mc(X, q). It can be seen that
the graphs are also good descriptions of the measured
values, and the above functions suggest that

q(nJ = 3 - 0,84 = 2,16 and

q(mc) = 6 - 3,80 = 2,20.

The two values obtained for q are in good agreement and,

being close to 2, indicate again that abrasive processes
were predominant in the mill. The mean value of q was
used in equation (30) for the calculation of
;\(q)

193 mm (7,6 inches).

This value of ;\ is in good agreement with those determined previously, which confirms the internal consistency
of the theory.
The mean of all the values of ;\ obtained so far gives
~Blyvooruitsig= 208 mm.

Quantitative values of a and {3in equation (17), the

expression for combined wear, can be determined for the
mill because the total number, total mass, and total rate
of consumption of the balls are known. When the mean
value of ;\ is substituted into equations (21), (24), and
(25), it is found (by use of the data in Table A-2 and
the assumption p = 7,6 X 103 kg m-3), that
1,97 X 10-5 t-t,
a
""
{3 ""6,5
X 10-5 kg1l3 Cl.

These values of a and (3permit the calculation of the

rate of abrasive wear, WA(m), and the rate of impactive
wear, W,(m), to which any ball of mass m in the mill

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

APRIL 1986

119

was subject. For example, for balls of average mass m

(0,558 kg),
WA(m) = (3m2/3 ::::::4,4

10-5 kg t-I,

and
Wj(m) = am ::::::1,1 x 10-5 kg cl.
These values, for balls of average mass, indicate that
abrasive interactions were predominant in the given mill.
That this was characteristic of the whole charge in the
mill can be confirmed by calculation of the rates of
abrasive wear and impactive wear on the whole ball
charge, viz WA(charge) and W1(charge). Quick estimates, which can be confirmed by detailed calculations
based on equations (27) and (28), indicate that, since Nm
is the total mass of the charge (13 309 kg), the charge
experienced an impactive wear rate of
Wj(charge

aNm

::::::0,26

kg t-I.

Hence the abrasive wear rate of the whole charge was

WA(charge) =

rate of ball consumption

W1(charge)
- 1,17 - 0,26
::::::0,91 kgt-l.

minus

The ratio RAI(charge) of the rates of abrasive to impactive wear on the ball charge was therefore
RAI(charge)

:::::: 3,5.

Substitution of A into equation (29) yields a value of 3,6

for RAI(charge), which is in good agreement with the
above.
The value of RAI(charge) clearly indicates that abrasion was probably the predominant comminution
mechanism in that mill and explains why Prenticel was
able to claim that the Blyvooruitsig data supported his
theory of ball wear. However, the results of the present
work show that about 23 per cent of the total ball consumption in the given mill was due to 'ID impactive
component.
The influence of the proportion vf non-spherical
grinding elements on the above results should be considered. The measured mass of the charge was 13 309 kg,
and the mass, as calculated from equation (24), was
12 334 kg, Le. the calculated mass was smaller than the
measured mass, as expected. The discrepancy of only 8
per cent is due to the non-sphericity of the smaller, worn
grinding elements in the charge. This suggests that the
error in the results for the abrasive component is about
82/3per cent, Le. 4 per cent, and that the impactive component is too high by about 8 per cent. If allowances are

made for uncertainties in 'A and Xo, it can be concluded

that the estimate of the ratio of the rates of abrasive to
impactive wear in the mill at Blyvooruitsig is, as far as
is known at present, reliable to about 20 per cent.
In the present work, ball samples were drawn from the
charges of the mills at the Libanon Gold Mine and the
Marikana Mine of Western Platinum Ltd. The data
120

APRIL 1986

describing the size distribution in the two ball samples

are given in Tables A- III and A- IV of the Addendum.
Fig. 3 compares the cumulative forms of the data (Le.
the number fractions of balls smaller than size X) relating
to these mills with curves representing the extremes of
purely abrasive wear and purely impactive wear, nA(X)
and nI(X) respectively. The data points are clearly intermediate between these two extremes, and the best
descriptions of the data are provided in each instance by
the graphs for the theory of combined wear, Le. the function nc(X,'A) as given by equation (25) in Table 11, with
the values
'ALibanon
= 149 mm
'AMarikana
= 10,2 mm.

These values of 'Aand the relevant data were substituted

into equation (29), and the ratio of the rates of abrasive
to impactive wear on the charge RAI(charge) was
calculated for both ball samples. The results (Table Ill)
show that the impactive component accounted for 74 and
34 per cent of the ball consumption in the Marikana and
Libanon mills respectively. Table III also shows that the
differences in the milling conditions at the two mines are
quite consistent with these results: the mill at Marikana
was equipped with lifter bars, which provided a strong
impactive component, whereas the mill at Libanon had
a grid liner, which was effectively smooth and which exhibited marked circumferential grooving due to slip and
therefore marked abrasion of the grinding charge. These
results are also consistent with those of Blyvooruitsig,
where the mill speed and diameter, volume of the charge
and top-size balls, and the imp active component (23 per
cent) were all less than at Libanon. The relationship between the magnitudes of the wear components and the
operating conditions of ball mills is an important corollary to the present work. The salient milling conditions
include the diameter and speed of the mill, the size of
the balls fed, the nature of the mill lining, and the pulp
density. It is contended that the relative magnitudes of
the wear components provide an indication of the
grinding mechanisms that are operative in ball mills, and
that the relative magnitudes can be changed by variation
of one or more of the salient milling conditions.
This matter was investigated further by the application
of the theory of combined wear to all the ball-size
distributions found in the literature, and the values of
q, 'A, and the relative magnitudes of the components in
the rate of ball wear were determined. The results of this
analysis are shown in Fig. 4, which was obtained by use
of the fact that 'A is a function of q and of the ball-size
reduction ratio, Xma/ Xo, as shown in equation (30). The
points in Fig. 4 show values of 'A/Xmaxfor the various
mills plotted as a function of the corresponding values
of q. The continuous curves in Fig. 4 show the theoretical
relationship-equation
(30)-between 'A/Xmaxand q for
various specific values of XmJ Xo' In spite of the limited
information and sometimes inadequate data (e.g. the size
range of the balls in four ball~tube mills at Lake Shore14
was divided into only three intervals), the measured values
exhibit a trend that provides strong confirmation of the
validity of the theory of combined wear proposed here.
At first sight, the measured values corresponding to ball

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

,-

(a) Libanon (A= 149mm)

Impactive

1,0

,/

"

component

(b) Marikana (A= 10,2 mm)

Impactive component 74%

34070

= Experimental
= nI(X); purely

data
impactive wear
= nA(X); purely abrasive wear
= nc(X, A); combined wear

'I

= Experimental

data

purely impactive wear

=nI(X);
nA(X); purely abrasive wear

= nc(X,A); combined wear

:'/

,/:/

:/
:'/
:/
:'/
:'I

0,8

1/

:/
:'/
:/
1/
:/
~/
:'I
:/
:'I
:'/
1// nA(X)
r4
.'/
:'/
:/
1/
:/
:'!
1/
:/

'"~

ci'
0

0,6

'~ ~
.:::

1;\
t't;;
oD bI)

EL5

a ~0. 04,
<1)

,:;

C;j
"'5

8
;::I

1/

0,2

/1
It

0
0

20

40

60

80

100

Ball size X, mm
Fig. 3-Comparison

40

60

Ball size X, mm

of the measured and calculated size distributions

charges at Sub Nigel1 and Hollinger1S appear to deviate

from the general trend, but this anomaly is resolved when
one notes that the ball scrap at these mills was recirculated
until it was completelyworn away. These values are clearly associated with curves appropriate to an infinite or very
large reduction ratio.
The majority of the points shown in Fig. 4 were
calculated from data reported by various authors, most
of whom interpreted their results in terms of either purely abrasive wear or purely imp active wear. The only exception was a guarded reference by Crockee4 to the
possibility of combined wear in the no. 7 ball mill at Lake
Shore. Fig. 4 shows that none of the data corresponds
to values of q equal to 2 or 3; on the contrary, the points
are fairly uniformly distributed between these two values.
Therefore, the present analysis and the physical evidence
discussed briefly in the introduction suggest strongly that
both wear mechanisms are present in all industrial mills.
In particular, although it was claimedl,7 that the ball
wear at Marievale7, HollingerlS, Blyvooruitsig1, and Sub
Nigel1 was due to purely abrasive interactions, the present re-analysis of the data confirmed that abrasion was
indeed predominant at Hollinger, Blyvooruitsig, and Sub
Nigel, but that imp active processes were predominant at

20

In samples of ball charges at Libanon and Marlkana Mines

Marievale. Information from the archives of General

Mining Union Corporation Limitedl7 has shown that, at
Marievale, the mill speed (more than 80 per cent of critical
speed), the volume of the ball charge (nearly 50 per cent
of the internal mill volume), and the top-size balls (4
inches) were consistent with the existence of an impactive component when the ball-size distribution in that mill
was measured.
Table III gives values of q, A, and the relative
magnitudes of the wear components for some of the mills.
Data for the four ball-tube mills at Lake Shorel4 were
ignored for the reasons given above. As indicated in Table
Ill, the gradual transition from predominantly impactive
to strongly abrasive conditions is consistent with accompanying changes in the conditions obtaining in the various
mills, and provides strong support for the contention that
the relative magnitudes of the wear components are
related to the milling conditions.
Summary and Conclusions
With the exception of Bond6,s, previous investigatorsl,3,7,Is,I9ascribed the wear of balls in milling to
only one wear mechanism. The most recent papero on
the subject, which apparently found 'the situation clear-

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

APRIL 1986

121

TABLE III

MILLING CONDITIONS

AND THE MAGNITUDE

Grinding balls

Mill

Diameter
ft

Mine

Length
ft

Speed
070
of
critical

Volume
of
charge
%

Top
size
mm

Miami CopperJ
(Hardinge conical
mill)

82

20

50,8

Marikana
Western Platinum
Limited

80

45

Marievale'

83

Lakeshorel4
no. 7 ball mill

Libanon

IN SOME INDUSTRIAL

0,0033

99,7

10,17

0,203

0,26

74

67,5

0,66

0,91

52

1,28

1,72

36

149

1,42

1,88

34,7

2,17

127

1,67

2,18

30,5

15

2,17

207

2,73

3,65

21,5

2,13

483

4,75

6,4

13,5

25

2,78

47

120

25

2,46

85

45

108

25

2,31

138

10

79

48

108

30

2,30

6Yz

12

80

50

76

Blyvoortuitsig'

6Y,

12

70

25

76

Sub Nigel'

6Yz

68

33

100

,
I

Sub
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

Nigel
i!)

Values determined from ball-size distribution data

Theoretical

relationship

at values

of Xm.,/

Xo:

-~2
~
~
-~

---

5
ID
00

I :

:\
:\
: \\
I

\1\:\\

~
>

11\

~
13
~

\\ \\\\

e~

\ \\
\ \\

E
0

to

Blyvo~r.

~\
\\
\.
\\ \\ \'.

'2

a.

~E

\ ~..Lakeshore2
\\

Hollinger1i>

~,

\ ...~ $

..
"

,\\o<e1

" \

):

~
0'< "
,..0;;:'
- - - \-', ""'..."
~'..;-~v~' -.~.. - - - - - ..~ - - -;;
'<-'...@'.,i.,o
~
:~;;.~..
"
\.,~"e

""

-"'

Fig.

2,0

2.2

4-Values

2,4

2,8

2.6

of AIXmlX as a function of q
distributions

122

""

APRIL 1986

~~~,g
.~ ~

~~~~~:~:~':;~~:.~~

"""
0

1i
'"

in a number

for the

of mills

0
u

~
3.0

bail-size

Ratio of Impactive
wear
abrasive
% of
0 impactAIXmax ive wear total wear
Small

60

15

BALL MILLS

Calculated parameters

2,97

7.5

Reject
size
mm

COMPONENT

Hollinger

~
):

OF THE IMPACTIVE

0,0047

ly confused', also subscribed to this view. The present

work has evidence that abrasive as well as provided impactive interactions occur among the grinding elements
in ball mills. A theory of combined wear was formulated
by the superposition of the wear rates of balls due to the
two types of interaction. New distribution functions based
upon this theory of combined wear were derived, and the
values determined from these functions were compared
with measurements of the ball-size distributions in a variety of industrial ball mills. The distribution of ball sizes
was shown to depend upon the relative magitudes of the
two wear mechanisms operating in ball mills, and the
relative magnitude of the two components in the rate of
ball wear were determined quantitatively for a large
number of industrial ball mills.
The relative intensities of the two wear mechanisms are
determined by the mechanical properties of the balls and
by such factors as the mill diameter and speed, the topsize balls added, the design of the mill lining, and the density of the mineral pulp. Hence, the magnitudes of the
components can be used as a basis for the selection of
the optimum chemical and metallurgical properties of
balls. They also provide indications of the mechanisms
of size reduction that are operating in any industrial ball
mill. This was confirmed by a comparison of measured
ball-size distributions with the milling conditions in a
number of industrial mills.
It was shown that the new formulae can be used to
determine the relative magnitudes of the wear components from the size distribution in a representative sample of balls from a mill. This obviates the laborious sizing
of the entire grinding charge of a ball mill. Rapid and

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

TABLE

III (cont.)

TABLE A-II
BALL-SIZE DISTRIBUTION AT THE BLYVOORUITSIG GOLD MINE,
18TH OCTOBER, 1942 (AFTER PRENTICE1)

Grading of
charge
inch

Remarks on milling conditions in relation to the

impactive and abrasive components
High-speed cataracting conditions, Iow solids content
in pUlpl6, and high ball-to-ball contact; hence impactive component almost totally predominant

<3
<2V,
<2
<1Y2
<I

High speed with lifter bars in linings; hence cataracting conditions and strong impactive component

>2Y2
>2
>1Y2
>1

Total

High speed and large grinding balls enhance impactive component, whereas smooth lining limits cataracting; hence impactive and abrasive components
almost equal

Mass
070

Average
Total mass mass per ball Number of
Ib
Ib
balls

53,34
2225
17,05
5,90
1,46

15 650
6528
5002
1732
428

3 114
1,774
0,890
0,333
0,099

5026
3680
5623
5198
4316

100,00

29 340

1,231-

23 843

-Average

Conditions similar to those at Marievale, but smaller

mill diameter increases abrasive component
to

Larger mill diameter, but slower speed tends to

increase abrasive component; circumferential grooving on the lining is evident

TABLE A-III
SIZE DISTRIBUTION IN A SAMPLE OF BALLSFROMTHE CHARGE OF
NO. 2 BALL MILL AT LIBANON GOLD MINE

Smaller mill diameter and smaller ball size increase

abrasive component

Size range
mm

Conditions similar to those at Hollinger but much

smaller ball charge increases abrasive component still
further

20
30
40
50
60
70
80
90
100

Mill of slowest speed gives highest abrasive

component

efficient methods for the taking of representative samples

would be of practical value.
The formulae and techniques of analysis developed in
the present work could also be useful in other applications where particles are consumed at rates proportional
to their dimensions, e.g. in chemical reactors, where the
dissolution of particles is a function, not only of their
exposed surfaces, but of the diffusion of the reactants
into the particles, i.e. a volume effect.

<4
<3Y2
<3
<2Y2
<2
<IV,
Total

>3Y2
>3
>2Y2
>2
>1Y2
>1

Number of
balls

45,3
18,7
17,6
7,7
8,6
2,1

2282
1423
2240
1792
4268
2935

100,0

14940

Prentice
theoryl

Davis
theory'

2185
2185
2 185
2 185
2 185
2 185

I 685
1947
2304
2820
3636
5124

13110

17 516

31
132
185
249
190
183
170
178
64
1382

TABLE A-IV
SIZE DISTRIBUTION IN A 344 kg SAMPLE OF BALLS FROM THE
CHARGE IN THE NO. 3A-STREAM BALL MILL AT THE MARIKANA
MINE OF WESTERN PLATINUM LIMITED

Ball size
mm

>30
>35
>40
>45
>50
>55

TABLE A-I
BALL-SIZE DISTRIBUTION AT MARIEVALE (AFTER WHITE')

Mass

30
40
50
60
70
80
90
100
105

Total

Addendum
Ball-size distribution data that were analysed by the
theory of combined wear are given in Tables A-I to
A-IV.

Ball size
inch

to
to
to
to
to
to
to
to
to

Number of
balls

<30
<35
<40
<45
<50
<55
<60

Total

Number of
balls
204
156
189
179
141
101
100
1070

Acknowledgements
This paper is published by permisison of the Council
for Mineral Technology (Mintek).
Thanks are due to Dr P.T. Wedephol for proposing
this investigation and to Dr M.A. Ford for suggesting fur-

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

APRIL 1986

123

ther extensions. Gratitude is also expressed to Professor

F.R.N. Nabarro, F.R.S., for discussions, criticism, and
comments, to Dr LA. Barker and Dr D. Hulbert for
discussions, to Mrs E. van der Berg, who assisted with
some computations, and to General Mining Union Corporation Limited, who made material from the Gencor
archives available. Discussions were also held with Mr
W. Flook of Gencor and Mr N.F. Peverett of Gold Fields
of South Africa Limited, to whom thanks are due for
the friendly interest they showed in this work. Grateful
acknowledgement is made of the assistance given by Mr
R.J. Adey and Mr C.L.M. Gough, the mill superintendents at Libanon Gold Mine and the Marikana Mine of
Western Platinum Limited respectively. The vigorous
cooperation of their milling personnel is also gratefully
acknowledged.
References
I. PRENTICE,T.K. Ball wear in cylindrical mills. J. Chem. Metall.
Min. Soc. S. Afr., Jan. 1943. pp. 99-116.
2. VERMEULEN,loA., and HOWAT, D.D. Abrasive and impactive
wear of high-chromium cast-iron grinding balls. Randburg, Council
for Mineral Technology, Report M116. 1984. 25 pp.
3. DAVIS,E.W. Fine crushing in ball mills. Trans. AIME, vol. 61.
1919. pp. 250-297.
4. VERMEULEN,loA., OHLSONDE FINE, M.J., and SCHAKOWSKI,
F.
Physical information from the inside of a rotary mill. J. S. Afr.
Inst. Min. Metall., vol. 84. 1984. pp. 247-253.
5. DuNN, D.J., and MARTIN,R.G. Measurement of impactive forces
in ball mills. Trans. Soc. Min. Eng. AIME, vol. 264. 1978. pp.

384-388.
6. BOND, F.C. Wear and size distribution
of grinding balls. Trans.
AIME,
vol. 153. 1943. pp. 373-384.
7. WHITE, H.A. Vote of thanks to the Prentice paper. J. Chem.
Metall. Min. Soc. S. Afr., Jan. 1943. pp. 116-122.
8. BOND, F.C. Written contribution
to discussion of the Prentice
paper. J. Chem. Metall. Soc. S. Afr., Jan-Feb.
1943. pp. 131-133.
9. HUKKI, R.T. Correlation
between principal parameters
affecting
mechanical ball wear. Trans. AIME, vol. 199. 1954. pp. 642-644.
10. TAGGART, A.F. Handbook
of mineral dressing. New York, John
Wiley & Sons, 1954. pp. 5-27.
11. VERMEULEN, loA., and How AT, D.D., Quantitative
assessment of
abrasive and impactive wear rates from ball-size distributions
in
rotary mills. Randburg,
Council for Mineral Technology,
Report
M201. May 1985.
12. BERNUTAT, P. Wear of grinding media and liner plates. ZementKalk Gips, no. 9. 1964. pp. 397-400.
13. ADAMSON, R.J. Gold metallurgy in South Africa. Johannesburg,
Chamber of Mines of South Africa, 1972. p. 45.
14. CROCKER, B.S. Written contribution
to the paper by T.K. Prentice. J. Chem. Metall. Min. Soc. S. Afr., Feb. 1944. pp. 133-136.
15. LONGMORE, E.lo, et al. Comparison
of Iow imd high discharge for
ball mills. Trans. Instn Min. Metall., vol. 46. 1937. pp. 562-583.
16. TAGGART, A.F. Op cit., p. 5-65.
17. FLOoK, W. General Mining Union Corporation
Ltd. Private communication,
1984.
18. NORMAN, T.E., and LOEB, CM. Wear tests on grinding balls.
Trans. AIME,
vol. 1983. 1949. pp. 330-360.
19. NORQUIST, D.E., and MOELLER, J.E. Relative wear rates of
various diameter grinding balls in production
mills. Trans. AIME,
vol. 187. 1950. pp. 712-714.
20. AuSTIN, loG., and KUMPEL, R.R. Ball wear and ball size distributions in tumbling ball mills. Power Technology,
vol. 41. 1985. pp.
279-286.

Ion-Ex '87
Papers are invited for Ion-Ex '87, an International
Conference and Industrial Exhibition on the industrial,
analytical,
and preparative
applications
of ion
chromatography and ion-exchange processes, which is to
be held on 13th to 16th April, 1987, in Wrexham, Wales.
The Conference is supported by the Royal Society of
Chemistry Analytical Division (North West Region) and
the Society of Chemical Industry, Solvent Extraction and
Ion Exchange Group, together with major organizations
involved in the field.
The proposed scope of the meeting is to include the
following general areas, and each will be reviewed by a
recognized authority.

. Inorganic ion analysis. Ion-exchangeresins, ion-

exchange processes and instrumentation (Review: Dr

Hamish Small)
Organic acid and base analysis,
including

.
.

biochemical, preparative, and assay techniques. Exchange resins and instrumentation (Review Dr F.C.
Smith, Millipore SA, France)
Polyelectrolyte fractionation processes
Industrial water-purification procedures, including
effluent treatment (Review: Dr J .R. Millar)

The proceedings of the Symposium will be published

by Elsevier Applied Science Publishers Ltd.
If you would like more information on Ion-Ex '87,
please contact
The North East Wales Institute
Connah's Quay
Deeside
Clwyd. CH5 4 BR
U.K.
Telephone 0244-817531 ext. 245. Telex: 61629
NEWI G.

IPMI conference
Dr M.L El Guindy has been appointed General Chairman of IPMI's 10th International Precious Metals Conference and Exhibition. The meeting is to be held at Lake
Tahoe, Nevada, from 9th to 12th June, 1986.
The theme of the Conference will be 'Interactive
Precious-Metal Technology-Producer
to User'. Sixtynine presentations will be given by international experts
on subjects such as precious metals in space-related industries, precious metals from natural resources, and
high- technology applications.
The Conference will mark the official celebration of
the tenth anniversary of the International Precious Metals
124

APRIL 1986

Institute. A special limited-edition one-ounce commemora' .ve silver medallion will be struck for the occasion. ~md a 52-page anniversary book highlighting the
history, accomplishments, and awards of IPMI will be
prepared.
Advance registration forms or additional information
can be obtained from
IPMI
Government Building
ABE Airport
Allentown, PA 18103,
U.S.A. Telephone: (215) 266-1570.

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

r~